Re: 4.9.0 regression in pipe-backed iov_iter with systemd-nspawn

[Al Viro]

On Fri, Jan 13, 2017 at 10:32:47AM +0000, Alan J. Wylie wrote:
> # dmesg  | tail
> ...
> [  141.553455] nr: 0->16, cur: 5->5, buffers: 16->16
> [  141.553461] copied: 0, count:2147479552, idx:5, offs:0

OK, that looks like pipe_advance() fencepost (hopefully) dealt with in
the lib/iov_iter.c patch I've just posted.  I'm going down right now, but
if you have a box you are not too fond of, it would be nice if you tried
that one.  Warning: the patch had been written in a barely awake state, and
is completely untested.  Treat accordingly...